Numerical control machining center protection method and system

Abstract

The invention relates to the technical field of numerical control machining centers, and provides a numerical control machining center protection method and system.The numerical control machining center protection method comprises the steps that vibration signals are collected and decomposed to obtain trend terms; a tool wear compensation factor is obtained based on the tool accumulated machining duration and the material removal amount, and compensation correction is conducted on the trend term according to the tool wear compensation factor; and obtaining a basic collision risk probability according to the compensated and corrected trend term. According to the method, the baseline offset of the vibration signal can be compensated and corrected in real time based on the accumulated machining time length of the cutter and the material removal amount, the time accumulation (thermal fatigue effect) and the cutting load (mechanical wear) of the cutter can be responded at the same time, the accuracy of basic collision risk probability calculation is improved, and safety protection of the numerical control machining center can be better achieved.

Description

Technical Field

[0001] This invention relates to the field of CNC machining center technology, and more specifically, to a protection method and system for CNC machining centers. Background Technology

[0002] A CNC machining center is a high-efficiency automated machine tool composed of mechanical equipment and a CNC system, suitable for machining complex parts. It is a CNC machine tool with a tool magazine that can automatically change tools and can perform a variety of machining operations within a certain range.

[0003] Chinese invention patent application number CN202410628428.4 discloses a data-driven method for monitoring the anti-collision machining status of CNC centers. It monitors the tool path during the machining process of CNC centers and calculates the real-time dynamic distance safety factor by combining the target position distance of the tool, the minimum safe distance from surrounding devices, speed, and acceleration data to determine whether there is an abnormal tool path that leads to a collision risk. It can more accurately detect the risk of tool collision from the actual movement of the tool.

[0004] As machining time increases, the normal current baseline rises, causing baseline shift and reducing the accuracy of tool collision risk calculation. The aforementioned data-driven CNC center anti-collision machining status monitoring method lacks a wear compensation correction mechanism based on tool machining time, further reducing the accuracy of tool collision risk calculation and leaving room for optimization. Summary of the Invention

[0005] Based on this, in order to address the problem that the accuracy of existing CNC machining center anti-collision monitoring methods needs to be improved in terms of tool collision risk, this invention provides a protection method and system for CNC machining centers, the specific technical solution of which is as follows:

[0006] A protection method for CNC machining centers includes the following steps:

[0007] Collect vibration signals and decompose the vibration signals to obtain trend terms;

[0008] The tool wear compensation factor is obtained based on the cumulative machining time and material removal amount, and the trend term is compensated and corrected according to the tool wear compensation factor.

[0009] The basic collision risk probability is obtained based on the trend term after compensation and correction.

[0010] The described protection method for CNC machining centers obtains a tool wear compensation factor by measuring the cumulative machining time and material removal amount, and compensates and corrects the trend term based on the tool wear compensation factor. It can compensate and correct the baseline offset of the vibration signal in real time based on the cumulative machining time and material removal amount, and can simultaneously respond to the tool's time accumulation (thermal fatigue effect) and cutting load (mechanical wear), improving the accuracy of the basic collision risk probability calculation and better realizing the safety protection of CNC machining centers.

[0011] Preferably, obtaining the basic collision risk probability based on the compensated and corrected trend term includes the following steps:

[0012] Acquire current signals, obtain current time series data of the current signals, and obtain the current time series standard deviation, current time series maximum value, and current time series minimum value based on the current time series data;

[0013] The current mutation factor is obtained based on the current time series standard deviation, the current time series maximum value, and the current time series minimum value.

[0014] Obtain the initial temperature distribution vector and the real-time temperature distribution vector of the tool, and obtain the thermal deformation coefficient based on the initial temperature distribution vector and the real-time temperature distribution vector;

[0015] The basic collision risk probability is obtained based on the current mutation factor, thermal deformation coefficient, and the compensated trend term.

[0016] Preferably, obtaining the basic collision risk probability based on the current mutation factor, thermal deformation coefficient, and compensated trend term includes the following steps:

[0017] The maximum amplitude of the spectral peak sequence of the vibration signal is obtained based on the compensated and corrected trend term;

[0018] Obtain the material resonance threshold, and obtain the resonance risk index based on the maximum amplitude and the material resonance threshold;

[0019] The basic collision risk probability is obtained based on the current mutation factor, thermal deformation coefficient, and resonance risk index.

[0020] Preferably, the protection method for CNC machining centers further includes the following steps:

[0021] The vibration signal is decomposed to obtain the seasonal term;

[0022] The basic collision risk probability is compensated and corrected based on the seasonal term to obtain the final collision risk probability.

[0023] Preferably, obtaining the tool wear compensation factor includes the following steps:

[0024] Obtain the tool reference life, and use the ratio of the cumulative machining time of the tool to the tool reference life as the time factor;

[0025] Obtain a preset critical removal volume, and use the ratio between the amount of material removed and the critical removal volume as the removal amount factor;

[0026] The tool wear compensation factor is obtained by weighting the time factor and the amount of material removed.

[0027] Preferably, the protection method for CNC machining centers further includes the following steps:

[0028] A multidimensional risk feature vector is constructed based on the current mutation factor, thermal deformation coefficient, and the compensated and corrected trend term.

[0029] Obtain the sensitivity adjustment factor, and map the product of the sensitivity adjustment factor and the multidimensional risk feature vector through the Sigmoid function to obtain the risk adjustment term;

[0030] The maximum possible displacement of the tool within a preset time period is obtained based on the tool's instantaneous feed rate, tool acceleration, and system control cycle.

[0031] Obtain the machine tool rigidity coefficient and the real-time Euclidean distance between the tool and the nearest obstacle;

[0032] The adaptive safety distance is obtained by the ratio between the product of the machine tool rigidity coefficient, the risk adjustment term, and the real-time Euclidean distance and the maximum possible displacement.

[0033] When the adaptive safety distance is less than a preset safety threshold, a preset avoidance strategy is executed.

[0034] A protection system for CNC machining centers, used to implement the aforementioned protection method for CNC machining centers, comprising:

[0035] The trend term acquisition module is used to collect vibration signals and decompose the vibration signals to obtain trend terms;

[0036] The wear compensation factor acquisition module is used to obtain the tool wear compensation factor based on the cumulative machining time of the tool and the amount of material removed.

[0037] The trend term compensation and correction module is used to compensate and correct the trend term according to the tool wear compensation factor.

[0038] The basic risk probability acquisition module is used to obtain the basic collision risk probability based on the trend term after compensation and correction.

[0039] Preferably, the basic risk probability acquisition module includes:

[0040] A current signal acquisition and processing unit is used to acquire current signals, obtain current time series data of the current signals, and obtain the current time series standard deviation, current time series maximum value, and current time series minimum value based on the current time series data.

[0041] The current mutation factor acquisition unit is used to acquire the current mutation factor based on the current time series standard deviation, the current time series maximum value, and the current time series minimum value.

[0042] A thermal deformation coefficient acquisition unit is used to acquire the initial temperature distribution vector and the real-time temperature distribution vector of the tool, and to acquire the thermal deformation coefficient based on the initial temperature distribution vector and the real-time temperature distribution vector.

[0043] The basic risk probability acquisition unit is used to acquire the basic collision risk probability based on the current mutation factor, thermal deformation coefficient, and the compensated and corrected trend term.

[0044] Preferably, the CNC machining center protection system further includes: a seasonal term acquisition module, used to decompose the vibration signal to acquire seasonal terms;

[0045] The final risk probability acquisition module is used to compensate and correct the basic collision risk probability based on the seasonal term to obtain the final collision risk probability.

[0046] Preferably, the CNC machining center protection system further includes:

[0047] The risk adjustment term acquisition module is used to acquire the sensitivity adjustment factor, construct a multidimensional risk feature vector based on the current mutation factor, thermal deformation coefficient and the compensated and corrected trend term, and map the product of the sensitivity adjustment factor and the multidimensional risk feature vector through the Sigmoid function to acquire the risk adjustment term.

[0048] The maximum possible displacement acquisition module is used to obtain the maximum possible displacement of the tool within a future preset time period based on the tool's instantaneous feed rate, tool acceleration, and system control cycle.

[0049] The real-time Euclidean distance acquisition module is used to acquire the machine tool rigidity coefficient and the real-time Euclidean distance between the tool and the nearest obstacle;

[0050] An adaptive safety distance acquisition module is used to acquire an adaptive safety distance based on the ratio between the product of the machine tool rigidity coefficient, the risk adjustment term, and the real-time Euclidean distance and the maximum possible displacement.

[0051] The avoidance strategy execution module is used to execute a preset avoidance strategy when the adaptive safety distance is less than a preset distance safety threshold. Attached Figure Description

[0052] Figure 1 This is a schematic diagram of the overall process of a protection method for a CNC machining center according to an embodiment of the present invention;

[0053] Figure 2 This is a flowchart illustrating a specific method for obtaining the tool wear compensation factor in one embodiment of the present invention;

[0054] Figure 3 This is a flowchart illustrating a specific method for obtaining the basic collision risk probability based on the compensated and corrected trend term in one embodiment of the present invention.

[0055] Figure 4 This is a flowchart illustrating a specific method for obtaining the basic collision risk probability in another embodiment of the present invention;

[0056] Figure 5 This is a schematic diagram of the overall process of a protection method for a CNC machining center according to another embodiment of the present invention. Figure 1 ;

[0057] Figure 6 This is a schematic diagram of the overall process of a protection method for a CNC machining center according to another embodiment of the present invention. Figure 2 ;

[0058] Figure 7 This is a schematic diagram of the overall structure of a protection system for a CNC machining center according to one embodiment of the present invention;

[0059] Figure 8 This is a schematic diagram of the functional module structure of the basic risk probability acquisition module in one embodiment of the present invention;

[0060] Figure 9 This is a schematic diagram of the overall structure of a CNC machining center protection system according to another embodiment of the present invention. Figure 1 ;

[0061] Figure 10 This is a schematic diagram of the overall structure of a CNC machining center protection system according to another embodiment of the present invention. Figure 2 ;

[0062] Figure 11 This is a schematic diagram of the workflow of a dynamic decision tree engine in one embodiment of the present invention;

[0063] Figure 12 This is a schematic diagram of the workflow of decomposing the original vibration signal using the RobustSTL algorithm in one embodiment of the present invention;

[0064] Figure 13 This is a schematic diagram illustrating the characteristic changes of a TC4 titanium alloy workpiece during processing simulation in one embodiment of the present invention;

[0065] Figure 14This is a schematic diagram comparing key indicators before and after compensation in one embodiment of the present invention;

[0066] Figure 15 This is a schematic diagram comparing the compensation effect during the tool life cycle in one embodiment of the present invention;

[0067] Figure 16 This is a comparison table of specific performance indicators before and after compensation in one embodiment of the present invention. Detailed Implementation

[0068] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to its embodiments.

[0069] Existing data-driven methods for monitoring collision avoidance machining status in CNC centers have the following problems: 1. They lack a wear compensation and correction mechanism based on tool machining time, reducing the accuracy of tool collision risk assessment. 2. They rely solely on servo motor current signals, ignoring multimodal data such as vibration frequency, and cannot capture non-current characteristic risks such as mechanical resonance, including latent collisions caused by tool chatter.

[0070] like Figure 1 As shown, a protection method for a CNC machining center according to one embodiment of the present invention includes the following steps:

[0071] S1. Acquire vibration signals and decompose the vibration signals to obtain trend terms. In signal decomposition, methods such as STL (Seasonal-Trend Decomposition using LOESS), EMD (Empirical Mode Decomposition), and wavelet transform can generally be used.

[0072] The trend term can be understood as the low-frequency component of the vibration signal, i.e., the original vibration signal Y(t) = T(t) + S(t) + R(t). The frequency of the trend term T(t) is usually below 10Hz, which is much lower than the dominant frequency of cutting vibration. It is used to reflect slow-changing processes such as tool wear, thermal deformation, and machine tool geometric errors, and can be used to monitor the remaining tool life. The frequency of the seasonal term S(t) is generally between 50-2000Hz, mainly used to reflect the periodic vibration of spindle rotation and gear meshing, and can be used for fault diagnosis (such as bearing damage). The residual term R(t) is generally greater than 2000Hz, manifesting as random impacts, measurement noise, etc. In general, the trend term is a quantitative indicator of system state degradation, and its amplitude growth slope is directly related to the tool health status.

[0073] For example, if the slope of the trend term is small, such as not exceeding 20% ​​of the warning threshold, it can be understood that the tool is in the initial grinding stage, with slight wear, and the system can maintain the corresponding process parameters; if the slope of the trend term increases slightly but is between 20% and 50% of the warning threshold, it can be understood that the tool is wearing steadily and thermal deformation is beginning to accumulate, and the system can increase the coolant flow rate for effective cooling; if the slope of the trend term is greater than the warning threshold, it indicates that the tool is severely worn and the stiffness exhibits non-linear decay, at which point speed reduction and tool change warning processing can be performed.

[0074] S2, based on the cumulative machining time and material removal amount, a tool wear compensation factor is obtained, and the trend term is compensated and corrected according to the tool wear compensation factor. As a preferred technical solution, such as... Figure 2 As shown, obtaining the tool wear compensation factor includes the following steps:

[0075] S21, obtain the tool reference life, and use the ratio between the cumulative machining time of the tool and the tool reference life as the time factor.

[0076] The tool reference life can be understood as the tool's service life under normal operating conditions. For example, the service life of a typical turning tool is 5000-8000 hours, that of an end mill is approximately 3000-4000 hours, that of a drill bit is 1000-2000 hours, and that of a reamer is approximately 500 hours. Of course, the reference life will vary depending on the material and type of the tool. The cumulative machining time can be recorded and obtained by the CNC system's timing module.

[0077] S22, obtain a preset critical removal volume, and use the ratio between the material removal amount and the critical removal volume as a removal amount factor. The material removal amount can be calculated in real time through cutting parameters. For example, the material removal amount is the cumulative sum of the products of the tool axial cutting depth, feed per tooth, cutting line speed, and sampling time interval over the time dimension. The preset critical removal volume can be based on experimental calibration or set by technicians based on production experience.

[0078] S23, obtain the tool wear compensation factor based on the weighted value of the time factor and the removal amount factor. For example, the tool wear compensation factor = time factor × first wear compensation coefficient + removal amount factor × second wear compensation coefficient. The first wear compensation coefficient and the second wear compensation coefficient can be set empirically or calibrated experimentally, and will not be elaborated further here.

[0079] Preferably, a time factor can be set first. The relevant power parameter β is used to obtain the material eigenvalue η of the tool. mThen, the power parameter is used as the power of the time factor to facilitate non-linear scaling of different time factors, and the removal factor V is adjusted. r / V0 performs a cube root calculation to mitigate abrupt changes under large cut-off conditions, avoiding over- or under-compensation due to linear volume superposition, thereby improving stability. The final tool wear compensation factor is expressed as: Among them, a w β w T represents the first wear compensation coefficient and the second wear compensation coefficient, respectively. c T0 and V represent the cumulative machining time and the tool reference life, respectively. r V0 and V0 represent the amount of material removed and the critical removal volume, respectively.

[0080] Generally, the power parameter β can be set to 1.0. Here, a method for dynamically adjusting the power parameter is provided: First, preset the tool wear levels of slight, moderate, and severe wear based on the cumulative machining time. The time factors corresponding to slight, moderate, and severe wear gradually increase. For example, the time factor ranges corresponding to slight, moderate, and severe wear can be set to 0-0.6, 0.6-0.8, and greater than 0.8 based on experience or experimental calibration. Then, the power parameter is adjusted in the form of a piecewise function according to the magnitude of the time factor, making the power parameter positively correlated with the tool wear level. For example, when the time factor range is 0-0.6, the power parameter can be set to 1.0; when the time factor range is 0.6-0.8, the power parameter can be set to 1.3; and when the time factor range is greater than 0.8, the power parameter can be set to 1.5 or 2.0. In this way, the time factor can be non-linearly amplified through the power parameter. The power parameter can also be set according to different tool materials to non-linearly amplify the time factor corresponding to tools of different materials.

[0081] The evolution of tool wear over time varies greatly across different machining scenarios. For example, in the machining of nickel-based superalloys, cutting heat causes the coating on the tool surface to peel off and the substrate to diffuse wear, with the wear rate accelerating over time (similar to an exponential law). In this case, the power parameter should be greater than 1 to amplify the decay effect over time. In the rough machining of titanium alloys, mechanical impact and adhesive wear are the main factors, and the wear increases approximately linearly over time. In this case, the power parameter is approximately equal to 1. In the milling of tenons and slots on aero-blades, the tool is subjected to periodic impacts, and the wear initially slows down and then accelerates over time (fatigue crack propagation). In this case, the power parameter should be optimized through experiments or machine learning to match the actual wear curve.

[0082] By setting this power exponent, the degree of nonlinear influence of cutting time on tool wear can be adjusted, adapting to the time-wear patterns under different materials and cutting conditions. This allows the model to adapt to the differentiated effects of materials (nickel-based alloys, titanium alloys, composite materials), tools (coatings, ceramics, PCD), and cutting parameters (speed, feed) on time wear, thereby improving the accuracy of the tool wear compensation factor.

[0083] The intrinsic coefficient of the material, which characterizes the resistance of the tool material to thermal degradation, can be obtained by first acquiring the fracture toughness, hardness, elastic modulus, and coefficient of thermal expansion of the tool material, and then calculating (fracture toughness × hardness) / (elastic modulus × coefficient of thermal expansion), or by obtaining it through laboratory calibration. For example, for tools made of cemented carbide, the intrinsic coefficient can be set to 1.0-1.2, and for tools made of ceramic materials, the intrinsic coefficient can be set to 0.8-0.9.

[0084] Assume the original trend term is represented as T. t The tool stiffness attenuation coefficient is denoted by K, and the trend term T after compensation and correction is... t * =T t -λ w ·K. Here, the tool stiffness attenuation coefficient is used to quantify the stiffness loss caused by tool wear, K∈[0,1] and K=0 for new tools and K=1 for fully worn tools, which can be calibrated experimentally.

[0085] In summary, this tool wear compensation factor responds to both time accumulation (thermal fatigue effect) and cutting load (mechanical wear), reflecting both the rapid decay of thermal-time coupling (such as high-speed cutting) and the gradual decay of force-volume coupling (such as large-mass milling), which is beneficial for achieving dynamic optimization of CNC machining parameters.

[0086] S3, obtain the basic collision risk probability based on the compensated and corrected trend term.

[0087] In the tool wear compensation factors, the time factor dominates thermochemical wear (such as diffusion wear and oxidation wear), while the removal amount factor dominates mechanical wear (such as abrasive wear and adhesive wear). The trend term is generated by the accumulation of thermal deformation due to tool stiffness decay, exhibiting a gradual change characteristic. By compensating and correcting the trend term through the tool wear compensation factors, firstly, it can compensate for the decrease in radial stiffness caused by tool wear, suppress tool deflection caused by cutting forces, thereby achieving stiffness decay correction; secondly, it can counteract the tool thermal elongation caused by frictional heating, solving the hole diameter taper problem in deep hole machining, thereby achieving thermal deformation neutralization.

[0088] For example, in step S3, the resonance risk index can be obtained first by compensating and correcting the trend term, and then the resonance risk index can be mapped to the 0-1 range by the Sigmoid function to obtain the basic collision risk probability.

[0089] After obtaining the basic collision risk probability, different strategies can be implemented and the process parameters of the CNC machining center can be adjusted according to the changes in the basic collision risk probability. For example, when the basic collision risk probability is greater than the early warning collision risk threshold, the feed speed can be appropriately reduced, or when the basic collision risk probability is greater than the alarm collision risk threshold, the machine can be stopped. The early warning collision risk threshold is less than the alarm collision risk threshold.

[0090] For example, when the basic collision risk probability is less than 0.6, the tool feed rate is reduced by 40%; when the basic collision risk probability is less than 0.8, the system performs radial tool retraction and coolant pressurization; when the basic collision risk probability is greater than 0.8, the system performs emergency stop and spindle brake.

[0091] In summary, the proposed protection method for CNC machining centers obtains a tool wear compensation factor by measuring the cumulative machining time and material removal amount, and then compensates and corrects the trend term based on the tool wear compensation factor. It can compensate and correct the baseline offset of the vibration signal in real time based on the cumulative machining time and material removal amount, and can simultaneously respond to the tool's time accumulation (thermal fatigue effect) and cutting load (mechanical wear), thereby improving the accuracy of the basic collision risk probability calculation and better realizing the safety protection of CNC machining centers.

[0092] As a preferred technical solution, in step S3, such as Figure 3 As shown, obtaining the basic collision risk probability based on the compensated and corrected trend term includes the following steps:

[0093] S31, acquire the current signal, obtain the current time series data of the current signal, and obtain the current time series standard deviation and the current time series maximum value I based on the current time series data. max and the minimum value of current timing I min .

[0094] S32, obtain the current mutation factor based on the current timing standard deviation, the current timing maximum value and the current timing minimum value.

[0095] Assume the current timing data is represented as I = [i1, i2, ..., i n The mean of the current time series data is expressed as μ. I Then the current time series standard deviation Current mutation factor Where, ε I =1×10-5 This is used to avoid the denominator being zero when the steady-state current (i.e., the maximum value of the current sequence) equals the minimum value of the current sequence, thus ensuring numerical stability.

[0096] The current mutation factor is mainly used to quantify the mutation intensity or abnormal fluctuation degree of the current signal. It can amplify the abnormal fluctuation signal and capture the abnormal state of the equipment by the ratio of the standard deviation to the dynamic range (the standard deviation of the molecular current time series is dominant). The influence of the operating condition can be offset by the normalization of the dynamic range, thereby improving the sensitivity and robustness.

[0097] S33, obtain the initial temperature distribution vector and the real-time temperature distribution vector of the tool, and obtain the thermal deformation coefficient based on the initial temperature distribution vector and the real-time temperature distribution vector.

[0098] Assume T' = [T'1, T'2, ..., T' p ] represents the initial temperature distribution vector, T” = [T” t_1 ,T” t_2 ,…,T” t_p Let ] represent the real-time temperature distribution vector at time t, and p represent the number of temperature measurement points. Then, the thermal deformation coefficient ΔT is the L2 norm of the difference between the initial temperature distribution vector and the real-time temperature distribution vector, specifically... This thermal distortion coefficient is mainly used to characterize the non-uniformity of thermal distribution. By calculating the L2 norm between vectors, it can better express the overall shift in temperature distribution.

[0099] S34, obtain the basic collision risk probability based on the current mutation factor, thermal deformation coefficient, and the compensated trend term, such as... Figure 4 As shown, the specific method for this step includes the following steps:

[0100] S341, obtain the maximum amplitude of the spectral peak sequence of the vibration signal based on the compensated and corrected trend term.

[0101] The original vibration signal Y(t) = T(t) + S(t) + R(t) contains a baseline offset due to factors such as tool wear and load accumulation in its trend term before compensation and correction. The maximum amplitude is artificially high at this point, which can easily lead to an artificially high resonance risk index. The trend term after compensation and correction is represented by T. t * =T t -λ w K eliminates baseline drift of amplitude caused by factors such as tool wear, allowing the maximum amplitude to more accurately reflect mechanical resonance energy.

[0102] S342, Obtain the material resonance threshold, and obtain the resonance risk index based on the ratio of the maximum amplitude to the material resonance threshold.

[0103] The resonance risk index is mainly used to detect the risk of mechanical resonance. The material resonance threshold can be understood as the critical acceleration at which the workpiece material resists resonance deformation. It can be calibrated experimentally. For example, the material resonance thresholds for titanium alloy TC4, aluminum alloy 7075, and high-temperature alloy GH4169 can be set to 8g, 5g, and 12g, respectively. Here, g is the acceleration due to gravity.

[0104] For example, the resonance risk index is equal to the maximum amplitude divided by the material resonance threshold. Different system decisions can be made for resonance risk indices of different ranges. Assuming the resonance risk index is normalized to the 0-1 range, when the resonance risk index is less than 0.6, the system decision is to stabilize processing; when the resonance risk index is between 0.6 and 0.8, the system decision is to set to a warning state; when the resonance risk index is greater than or equal to 0.8, the system decision is to set to flutter approximation, and when the resonance risk index is greater than 0.9, a danger alarm signal and emergency shutdown are triggered.

[0105] Assuming the workpiece to be machined is a thin-walled titanium alloy structural component made of TC4 material, with a material resonance threshold of 8g, the traditional method, due to the lack of tool wear compensation, will cause the baseline vibration amplitude to rise by 1.2g after 3 hours of machining, reaching a maximum amplitude of 8.3g. At this point, the resonance risk threshold = 8.3 / 8 = 1.04, leading to false alarms and triggering an emergency shutdown. After compensating for tool wear, i.e., reducing the maximum amplitude of the vibration signal's spectral peak sequence from 8.3g to 7.1g based on the compensated trend term, the resonance risk threshold = 7.1 / 8 = 0.89. In this case, the system decision is set to chatter approximation, and no danger alarm signal or emergency shutdown will be triggered.

[0106] In summary, the resonance risk threshold can be dynamically adjusted by compensating for the corrected trend term. Based on this resonance risk threshold, different system decisions can be made (e.g., automatically reducing spindle speed to avoid the resonance frequency band when the resonance risk threshold is too high), which can reduce the system false alarm rate and machine tool structural fatigue damage, providing highly reliable protection for CNC machining.

[0107] S343, obtain the basic collision risk probability based on the current mutation factor, thermal deformation coefficient and resonance risk index.

[0108] For example, the basic collision risk probability P can be obtained by mapping the weighted values ​​of the current mutation factor, thermal deformation coefficient, and resonance risk index using the Sigmoid function. base Specifically, P base =σ(ω1C v +ω2R r +ω3ΔT), where ω1, ω2, and ω3 represent the proportional coefficients of the current mutation factor, thermal deformation coefficient, and resonance risk index, respectively, and σ(·) represents the Sigmoid function.

[0109] In this embodiment, the weighted values ​​of the current mutation factor, thermal deformation coefficient, and resonance risk index are mapped by the Sigmoid function to obtain the basic collision risk probability. It integrates multi-modal signals including current sequence, mechanical vibration spectrum peak sequence, and thermal field (temperature distribution vector), which breaks through the limitation of single signal and overcomes the defects of the binary judgment of risk probability in the prior art, and can quantify the degree of collision risk level.

[0110] As a preferred technical solution, such as Figure 5 As shown, the protection method for CNC machining centers further includes the following steps:

[0111] S4, decompose the vibration signal to obtain the seasonal term.

[0112] S5, the basic collision risk probability is compensated and corrected according to the seasonal term to obtain the final collision risk probability.

[0113] like Figure 12 As shown, this embodiment uses the RobustSTL algorithm to decompose the original vibration signal and obtain the trend term, seasonal term, and residual term.

[0114] The seasonal term S(t) is generally composed of the spectral amplitude sequence of the machining periodic vibration reflecting the intensity of chip force fluctuations, the fundamental frequency, and the phase angle. Its characteristic parameters mainly include the dominant frequency amplitude, phase shift, and harmonic energy ratio. Compensating for the basic collision risk probability based on the seasonal term can be understood as obtaining a seasonal correction factor based on the seasonal term, and then compensating for the basic collision risk probability based on this seasonal correction factor. The final collision risk probability obtained is: basic collision risk probability × seasonal correction factor. Here, the seasonal correction factor can be expressed as a piecewise function, described as follows: when the maximum value of the spectral amplitude sequence of the machining periodic vibration is greater than a preset safety threshold, the seasonal correction factor is set to the first seasonal correction value, such as 1.2; when the phase angle is abnormal (phase shift of 10 or 15 degrees), the seasonal correction factor is set to the second seasonal correction value, such as 0.8 × e -||Δφ|| Δφ represents the phase shift, e is the natural constant, and ||Δφ|| represents the norm of the phase shift. When the harmonic imbalance, such as the harmonic energy ratio, is greater than 0.25 or 0.3, the seasonal correction factor is set to the third seasonal correction value, such as 1 + 0.5 × harmonic energy ratio.

[0115] In traditional CNC machine tools, seasonal terms are generally only used for filtering. This embodiment dynamically corrects the basic collision risk probability using seasonal terms. By utilizing the phase shift characteristics of seasonal terms, compensation can be triggered in the early stages of tool wear, and the harmonic energy ratio can effectively distinguish between real collisions and electromagnetic interference.

[0116] The residual term = original vibration signal - compensated trend term - seasonal term, which can be understood as the transient abnormal component after stripping away trend and periodic characteristics. Residual term characteristics generally include sudden increases in pulse amplitude, high-frequency energy accumulation, and non-Gaussian distribution shift. Sudden increases in pulse amplitude are often caused by tool chipping / workpiece hard point impact, high-frequency energy accumulation is caused by instantaneous vibration caused by chip entanglement, and non-Gaussian distribution shift is caused by bearing failure / local jamming of the guide rail.

[0117] Here, the frequency band energy entropy E can be obtained by first decomposing the residual terms using wavelet packets. f and kurtosis coefficient K u Then, the basic collision risk probability is compensated and corrected based on the seasonal term and the residual term to obtain the final collision risk probability = basic collision risk probability × seasonal correction factor + residual correction value. First, the residual correction value is obtained based on the residual term, and this residual correction value is expressed as a piecewise function.

[0118] For example, residual correction value Where H represents the Hurst exponent of the residual term sequence, τ impat This represents the impact toughness threshold of the material, which can be determined experimentally or set empirically; for example, it can be set to 150 g / ms for titanium alloys. max|R(t)| represents the maximum value among the absolute values ​​of the residual term sequence, ||▽R(t)||2 represents the gradient norm of the residual term, and β represents the material fatigue coefficient.

[0119] When the kurtosis coefficient is greater than 3.5, it can be understood as an impact-dominated scenario, and 0.25 is an empirical factor obtained through experimental calibration. When the frequency band energy entropy is greater than 2.0, it can be understood as an energy accumulation scenario, and 0.15 is a material degradation sensitivity factor obtained through experimental calibration. The continuous micro-impact scenario refers to transient impact events that occur repeatedly in a mechanical system, with weak amplitude but significant cumulative effects. These impacts are easily masked by noise due to their small energy, but high-frequency accumulation can cause material fatigue damage. The continuous micro-impact scenario can be understood as a high-frequency impact with a single impact force less than 5% of the system's rated load, a single impact duration less than 1 millisecond, and a frequency greater than 1000 times / minute. Based on the changes in the residual term characteristics, it corresponds to different processing events and system response strategies, as detailed in the table below:

[0120]

[0121] It should be noted that the parameters in the piecewise function of the residual correction value and the relevant parameters in the table above can be adjusted according to the actual situation. By integrating the seasonal term and the residual term to compensate and correct the basic collision risk probability, the final collision risk probability is obtained. This takes into account the sudden risks dominated by the instantaneous rate of change (such as chipping) and the gradual faults controlled by energy integral (such as bearing wear), which can further improve the accuracy of the collision risk probability of CNC machine tools.

[0122] As a preferred technical solution, such as Figure 6 As shown, the protection method for CNC machining centers further includes the following steps:

[0123] S6. Construct a multi-dimensional risk feature vector based on the current mutation factor, thermal deformation coefficient, and the compensated and corrected trend term. Specifically, first obtain the resonance risk index based on the compensated and corrected trend term, and then construct the multi-dimensional risk feature vector based on the current mutation factor, thermal deformation coefficient, and resonance risk index.

[0124] S7. Obtain the sensitivity adjustment factor. Map the product of the sensitivity adjustment factor and the multidimensional risk feature vector using the Sigmoid function to obtain the risk adjustment term. This sensitivity adjustment factor can be set according to different processing scenarios or dynamically optimized based on historical collision data through online learning to adjust the sensitivity of the risk adjustment term.

[0125] Preferably, the sensitivity adjustment factor can be dynamically optimized based on the effectiveness of historical avoidance strategies using an online parameter tuning strategy based on reinforcement learning, so as to achieve adaptive security protection and balance the false alarm rate (overprotection) and the false negative rate (underprotection).

[0126] Specifically, k t+1 =k t +α'·(1-e -β'·SR ), where k t+1 k t α and β represent the sensitivity adjustment factors at time t+1 and time t, respectively. α' and β' represent the base learning rate and convergence control coefficient, respectively. The base learning rate controls the adjustment range of the sensitivity adjustment factor and is an empirical value, generally between 0.05 and 0.3. The base learning rate often varies depending on the industrial scenario. For example, in high-speed finishing, its value is between 0.05 and 0.1 to achieve fine-tuning of the sensitivity adjustment factor and avoid frequent process disturbances. In heavy-duty roughing, the base learning rate is between 0.2 and 0.3 to quickly respond to sudden changes in risk. In flexible assembly lines, the base learning rate is around 0.15 to balance efficiency and safety.

[0127] The convergence control coefficient is used to adjust the non-linearity of the learning speed. Decreasing its value prevents excessive reduction of the sensitivity adjustment factor due to occasional failures, while increasing it accelerates the increase of the sensitivity adjustment factor to strengthen protection. For example, if SR is less than 0.7, the convergence control coefficient can be set to 0.5; if SR is greater than or equal to 0.9, the convergence control coefficient can be set to 2.0. SR represents the effectiveness of historical avoidance strategies, which is equal to the number of effective avoidances in the last N policy executions divided by the total number of triggers.

[0128] The sensitivity adjustment factor function model is dynamically optimized and updated based on the effectiveness of historical avoidance strategies. This allows it to adapt to different operating conditions, avoid over-response, and drive the sensitivity adjustment factor towards the optimal solution. Compared to fixed sensitivity adjustment factor function models that require periodic calibration, are prone to failure due to equipment aging, and require recalibration when crossing scenarios, the sensitivity adjustment factor function model in this embodiment features adaptive optimization updates, automatic transfer learning, and continuous self-evolution.

[0129] S8, based on the instantaneous feed rate v of the tool t Tool acceleration a t The system control cycle is used to obtain the maximum possible displacement of the tool within a preset future time period. Assuming the system control cycle is denoted as Δt (typically 2 milliseconds), the maximum possible displacement within the preset future time period Δt is...

[0130] S9 obtains the machine tool rigidity coefficient and the real-time Euclidean distance between the tool and the nearest obstacle.

[0131] S10, an adaptive safety distance is obtained based on the ratio between the product of the machine tool rigidity coefficient, the risk adjustment term, and the real-time Euclidean distance and the maximum possible displacement.

[0132] The machine tool rigidity coefficient characterizes the impact resistance of the machine tool structure and is determined by the bed material and mechanical design. Generally, it is directly proportional to the product of the bed's elastic modulus and the moment of inertia of the machine tool frame section, and inversely proportional to the spindle overhang length. For precision engraving and milling machines, the machine tool rigidity coefficient is between 0.55 and 0.75, and for five-axis gantry milling machines, the machine tool rigidity coefficient is between 1.2 and 1.8.

[0133] Compared to the fixed safety threshold of traditional CNC machining centers, the adaptive safety distance integrates multimodal risk signals for dynamic adjustment, taking into account the machine tool rigidity coefficient, risk adjustment term, real-time Euclidean distance and maximum possible displacement. Through three-dimensional coupling of position, motion and risk, it has spatial perception and motion prediction functions.

[0134] S11, when the adaptive safety distance is less than the preset safety threshold, the preset avoidance strategy is dynamically executed, thereby improving the reliability of the safety protection of the CNC machining center.

[0135] The preset avoidance strategies include, but are not limited to, adjusting the spindle speed, path offset angle, and cutting force. For example, assuming the actual tool-fixture distance is 1.0mm, during the machining of thin-walled parts, if the adaptive safety distance decreases from 1.2mm to 0.8mm due to the increase in vibration amplitude, and the final collision risk probability is greater than the alarm collision risk threshold, then an avoidance strategy of appropriately reducing the spindle speed and cutting force can be adopted until the adaptive safety distance recovers to 1.2mm and the final collision risk probability is less than the alarm collision risk threshold.

[0136] like Figure 7 As shown, a CNC machining center protection system according to one embodiment of the present invention is used to implement the CNC machining center protection method, including a trend term acquisition module, a wear compensation factor acquisition module, a trend term compensation correction module, and a basic risk probability acquisition module.

[0137] The trend term acquisition module collects vibration signals and decomposes them to obtain trend terms; the wear compensation factor acquisition module obtains a tool wear compensation factor based on the cumulative machining time and material removal amount; the trend term compensation and correction module compensates and corrects the trend terms according to the tool wear compensation factor; and the basic risk probability acquisition module obtains the basic collision risk probability based on the compensated and corrected trend terms. Here, the resonance risk index can be obtained first through the compensated and corrected trend terms, and then the resonance risk index can be mapped to the 0-1 range using the Sigmoid function to obtain the basic collision risk probability.

[0138] Preferably, such as Figure 8 As shown, the basic risk probability acquisition module includes a current signal acquisition and processing unit, a current mutation factor acquisition unit, a thermal deformation coefficient acquisition unit, and a basic risk probability acquisition unit.

[0139] The current signal acquisition and processing unit is used to acquire current signals, obtain current time-series data of the current signals, and obtain the current time-series standard deviation, maximum current time-series value, and minimum current time-series value based on the current time-series data. The current mutation factor acquisition unit is used to obtain the current mutation factor based on the current time-series standard deviation, maximum current time-series value, and minimum current time-series value. The thermal deformation coefficient acquisition unit is used to acquire the initial temperature distribution vector and real-time temperature distribution vector of the tool, and obtain the thermal deformation coefficient based on the initial temperature distribution vector and real-time temperature distribution vector. The basic risk probability acquisition unit is used to obtain the basic collision risk probability based on the current mutation factor, thermal deformation coefficient, and the compensated and corrected trend term.

[0140] For example, the resonance risk index can be obtained first by compensating and correcting the trend term, and then the weighted values ​​of the current mutation factor, thermal deformation coefficient, and resonance risk index can be mapped by the Sigmoid function to obtain the basic collision risk probability P. base .

[0141] In this embodiment, the weighted values ​​of the current mutation factor, thermal deformation coefficient, and resonance risk index are mapped by the Sigmoid function to obtain the basic collision risk probability. It integrates multi-modal signals including current sequence, mechanical vibration spectrum peak sequence, and thermal field (temperature distribution vector), which breaks through the limitation of single signal and overcomes the defects of the binary judgment of risk probability in the prior art, and can quantify the degree of collision risk level.

[0142] As a preferred technical solution, such as Figure 9 As shown, the CNC machining center protection system further includes a risk adjustment item acquisition module, a maximum possible displacement acquisition module, a real-time Euclidean distance acquisition module, an adaptive safety distance acquisition module, and an avoidance strategy execution module.

[0143] The risk adjustment term acquisition module is used to obtain a sensitivity adjustment factor. It constructs a multi-dimensional risk feature vector based on the current mutation factor, thermal deformation coefficient, and the compensated trend term. The product of the sensitivity adjustment factor and the multi-dimensional risk feature vector is mapped using a Sigmoid function to obtain the risk adjustment term. The sensitivity adjustment factor can be set according to different processing scenarios or dynamically optimized based on historical collision data through online learning to adjust the sensitivity of the risk adjustment term. The risk adjustment term quantifies the multi-dimensional risk feature vector using a Sigmoid function, mapping complex risk factors into continuous risk probabilities that can participate in adaptive safety distance calculations, replacing traditional binary judgment.

[0144] The maximum possible displacement acquisition module is used to obtain the maximum possible displacement of the tool within a preset time period in the future based on the tool's instantaneous feed rate, tool acceleration, and system control cycle; the real-time Euclidean distance acquisition module is used to obtain the machine tool rigidity coefficient and the real-time Euclidean distance between the tool and the nearest obstacle.

[0145] Assuming the system control period is Δt (typically 2 milliseconds), the maximum possible displacement within the preset time period Δt is: The larger the maximum possible displacement, the more "aggressive" the tool movement (higher speed / acceleration), the higher the "risk exposure level" per unit spatial distance, the wider the area the tool "may reach" in the near future, the higher the collision risk, and the smaller the required safety distance. In other words, the adaptive safety distance for triggering a safety response must be more stringent. Conversely, when the tool moves at low speed, the value of this maximum possible displacement is relatively small, allowing for closer proximity, and the adaptive safety distance can be relatively larger.

[0146] The machine tool rigidity coefficient reflects the equipment's resistance to deformation and vibration. A smaller value indicates a more susceptible machine tool to deformation and vibration under external forces (such as potential collision forces when approaching obstacles), thus requiring greater safety redundancy. Conversely, a larger machine tool rigidity coefficient results in higher equipment stability, allowing for a more appropriate reduction in safety redundancy. Therefore, the adaptive safety distance is directly proportional to the machine tool rigidity coefficient; a smaller rigidity coefficient necessitates a more stringent adaptive safety distance for triggering a safety response.

[0147] The adaptive safety distance acquisition module is used to obtain the adaptive safety distance based on the ratio between the product of the machine tool rigidity coefficient, the risk adjustment term, and the real-time Euclidean distance and the maximum possible displacement. The avoidance strategy execution module is used to execute a preset avoidance strategy when the adaptive safety distance is less than a preset safety threshold. Specifically, the adaptive safety distance = (machine tool rigidity coefficient × risk adjustment term × real-time Euclidean distance) / maximum possible displacement. The working mechanism and underlying logic of the adaptive safety distance are explained below:

[0148] 1. The real-time Euclidean distance is the minimum Euclidean distance between the tool and obstacle position vectors, accurately representing the most dangerous (closest) spatial relationship between the two, and is the basic spatial metric for adaptive safety distance calculation. The maximum possible displacement describes the tool's motion displacement within a time interval Δt. It is based on the kinematic formula (composite of uniform velocity and uniform acceleration displacement), reflects the dynamic motion characteristics of the equipment, and reflects the trend of tool position change in the near future.

[0149] 2. The machine tool rigidity coefficient, as an inherent parameter of CNC machining centers, quantifies the impact of machine tool physical characteristics (such as structural stiffness and vibration suppression capability) on safety distances. Machine tools with poor rigidity require greater safety redundancy, which can be determined through equipment calibration or experimental testing. The sensitivity adjustment factor is a dynamic parameter that can be updated through online learning, controlling the "steepness" of the risk adjustment term curve. The larger the value, the more sensitive the risk response; that is, even small changes in risk can significantly alter the safety distance, adapting to the risk tolerance requirements of different processes.

[0150] 3. The adaptive safety distance incorporates the maximum possible displacement into the device's motion state, allowing the adaptive safety distance to change in real time with motion parameters. In high-speed / high-acceleration scenarios, the maximum possible displacement increases, and the safety distance is dynamically adjusted to cover a "farther potential collision range," which aligns with the logic in actual processing that "the more intense the motion, the higher the safety redundancy requirement."

[0151] 4. Traditional threshold functions use a binary "collision / safety" judgment. The risk adjustment term described here transforms risk into a continuous probability through the Sigmoid function, overcoming the limitation of a black-and-white approach. The multi-dimensional risk feature vector integrates multiple factors and adjusts the risk response sensitivity through a sensitivity adjustment factor, allowing the safety distance to subtly reflect the "degree of danger" (e.g., a slight increase in safety distance for minor collision risks and a significant increase for high risks), better reflecting the dynamic risk characteristics of complex processing scenarios.

[0152] 5. The machine tool rigidity coefficient is associated with the machine tool rigidity (physical property), the real-time Euclidean distance is associated with the spatial position (geometric constraint), the maximum possible displacement is associated with the control characteristics (dynamic motion), and the risk adjustment term quantifies the multi-dimensional risk feature vector through the Sigmoid function, thereby constructing a complete logical chain of "physical constraint → dynamic motion → risk decision", systematically ensuring machining safety from the essence of the equipment to the control requirements.

[0153] The preset avoidance strategies include, but are not limited to, adjusting the spindle speed, path offset angle, and cutting force. For example, assuming the actual tool-fixture distance is 1.0mm, during the machining of thin-walled parts, if the adaptive safety distance decreases from 1.2mm to 0.8mm due to the increase in vibration amplitude, and the final collision risk probability is greater than the alarm collision risk threshold, then an avoidance strategy of appropriately reducing the spindle speed and cutting force can be adopted until the adaptive safety distance recovers to 1.2mm and the final collision risk probability is less than the alarm collision risk threshold.

[0154] Compared to the fixed safety threshold of traditional CNC machining centers, the adaptive safety distance integrates multimodal risk signals for dynamic adjustment, taking into account the machine tool rigidity coefficient, risk adjustment term, real-time Euclidean distance and maximum possible displacement. Through three-dimensional coupling of position, motion and risk, it has spatial perception and motion prediction functions.

[0155] As a preferred technical solution, such as Figure 10 As shown, the CNC machining center protection system further includes a seasonal item acquisition module and a final risk probability acquisition module. The seasonal item acquisition module is used to decompose the vibration signal to obtain seasonal items, and the final risk probability acquisition module is used to compensate and correct the basic collision risk probability based on the seasonal items to obtain the final collision risk probability.

[0156] In traditional CNC machine tools, seasonal terms are generally only used for filtering. This invention dynamically corrects the basic collision risk probability through seasonal terms, utilizes the phase shift characteristics of seasonal terms to trigger compensation in the early stages of tool wear, and effectively distinguishes between real collisions and electromagnetic interference through harmonic energy ratio.

[0157] As a preferred technical solution, such as Figure 11 As shown, the CNC machining center protection system also includes a dynamic decision tree engine. The dynamic decision tree engine takes a multi-dimensional risk feature vector as input, realizes branching logic through feature comparison and node type judgment, and finally outputs the collision risk probability and pattern strategy.

[0158] Here, the left subtree is used to execute low-risk decision paths, pointing to a more conservative decision space (such as deceleration or increasing the safety distance). Specifically, the left subtree mainly handles low-risk situations and outputs conservative optimization strategies. The right subtree is used to execute high-risk decision paths, pointing to aggressive avoidance strategies (such as emergency stops or path replanning). Specifically, the right subtree mainly handles high-risk situations and triggers active protection mechanisms. Based on this dynamic decision tree engine, it uses a multi-dimensional risk feature vector composed of multi-modal signals as input, which can overcome the limitations of a single signal and ultimately obtain reliable risk probabilities and avoidance path suggestions, solving the rigidity problem of traditional isolated forest algorithms.

[0159] In one embodiment of the present invention, the CNC machining center protection system of the present invention also performs machining simulation on a TC4 titanium alloy workpiece. The simulation data of the machining process includes characteristic changes in three wear stages. In the initial wear stage (0-20 min), the tool wear compensation factor is mainly affected by the cube root term. The time factor is dominant, reflecting the characteristics of mechanical wear. In the stable wear stage (20-100 min), the time factor and the removal amount factor reach equilibrium, and the compensation amount increases linearly. In the severe wear stage (>100 min), the exponential effect of the time factor becomes apparent, and the system automatically triggers a tool change warning (tool wear compensation factor greater than 0.9). For example... Figure 13 As shown, with the increase of machining time and the deepening of tool wear, the characteristics such as spindle power, tool infrared temperature and dimensional error gradually increase, which is consistent with the actual machining situation.

[0160] like Figure 14 , Figure 15 as well as Figure 16 As shown, after compensating and correcting the trend term based on the tool wear compensation factor, obtaining the basic collision risk probability, and implementing the corresponding avoidance strategy, key indicators, including dimensional error, surface roughness, and tool life, have all been significantly improved. Specifically, the dimensional error before compensation was ±0.12mm, and after compensation, it was ±0.03mm, an improvement of 75% in accuracy. The surface roughness before compensation was 1.8μm, and after compensation, it was 0.6μm, an improvement of 200% in quality. The tool replacement frequency before compensation was 2 pieces / insert, and after compensation, it was 5 pieces / insert, a reduction of 60% in cost.

[0161] This simulation verification shows that the dual physical quantity (cumulative tool processing time and material removal amount) coupled compensation model of the CNC machining center protection method and system can improve the process stability in titanium alloy processing compared with the traditional single-dimensional compensation method, and is particularly suitable for the processing of precision parts such as aero-engine blades.

[0162] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. A protection method for a CNC machining center, characterized in that, Includes the following steps: Collect vibration signals and decompose the vibration signals to obtain trend terms; The tool wear compensation factor is obtained based on the cumulative machining time and material removal amount, and the trend term is compensated and corrected according to the tool wear compensation factor. The basic collision risk probability is obtained based on the trend term after compensation and correction.

2. The protection method for a CNC machining center as described in claim 1, characterized in that, Obtaining the basic collision risk probability based on the compensated and corrected trend term includes the following steps: Acquire current signals, obtain current time series data of the current signals, and obtain the current time series standard deviation, current time series maximum value, and current time series minimum value based on the current time series data; The current mutation factor is obtained based on the current time series standard deviation, the current time series maximum value, and the current time series minimum value. Obtain the initial temperature distribution vector and the real-time temperature distribution vector of the tool, and obtain the thermal deformation coefficient based on the initial temperature distribution vector and the real-time temperature distribution vector; The basic collision risk probability is obtained based on the current mutation factor, thermal deformation coefficient, and the compensated trend term.

3. The protection method for a CNC machining center as described in claim 2, characterized in that, The basic collision risk probability is obtained based on the current mutation factor, thermal deformation coefficient, and compensated trend term, including the following steps: The maximum amplitude of the peak spectral sequence of the vibration signal is obtained from the compensated and corrected trend term. Obtain the material resonance threshold, and obtain the resonance risk index based on the ratio of the maximum amplitude to the material resonance threshold; The basic collision risk probability is obtained based on the current mutation factor, thermal deformation coefficient, and resonance risk index.

4. The protection method for a CNC machining center as described in claim 3, characterized in that, It also includes the following steps: The vibration signal is decomposed to obtain the seasonal term; The basic collision risk probability is compensated and corrected based on the seasonal term to obtain the final collision risk probability.

5. A protection method for a CNC machining center as described in claim 4, characterized in that, Obtaining the tool wear compensation factor involves the following steps: Obtain the tool reference life, and use the ratio of the cumulative machining time of the tool to the tool reference life as the time factor; Obtain a preset critical removal volume, and use the ratio between the amount of material removed and the critical removal volume as the removal amount factor; The tool wear compensation factor is obtained by weighting the time factor and the amount of material removed.

6. The protection method for a CNC machining center as described in claim 5, characterized in that, It also includes the following steps: A multidimensional risk feature vector is constructed based on the current mutation factor, thermal deformation coefficient, and the compensated and corrected trend term. Obtain the sensitivity adjustment factor, and map the product of the sensitivity adjustment factor and the multidimensional risk feature vector through the Sigmoid function to obtain the risk adjustment term; The maximum possible displacement of the tool within a preset time period is obtained based on the tool's instantaneous feed rate, tool acceleration, and system control cycle. Obtain the machine tool rigidity coefficient and the real-time Euclidean distance between the tool and the nearest obstacle; The adaptive safety distance is obtained by the ratio between the product of the machine tool rigidity coefficient, the risk adjustment term, and the real-time Euclidean distance and the maximum possible displacement. When the adaptive safety distance is less than a preset safety threshold, a preset avoidance strategy is executed.

7. A protection system for a CNC machining center, used to implement the protection method for a CNC machining center as described in any one of claims 1-6, characterized in that, include: The trend term acquisition module is used to collect vibration signals and decompose the vibration signals to obtain trend terms; The wear compensation factor acquisition module is used to obtain the tool wear compensation factor based on the cumulative machining time of the tool and the amount of material removed. The trend term compensation and correction module is used to compensate and correct the trend term according to the tool wear compensation factor. The basic risk probability acquisition module is used to obtain the basic collision risk probability based on the trend term after compensation and correction.

8. The CNC machining center protection system as described in claim 7, characterized in that, The basic risk probability acquisition module includes: A current signal acquisition and processing unit is used to acquire current signals, obtain current time series data of the current signals, and obtain the current time series standard deviation, current time series maximum value, and current time series minimum value based on the current time series data. The current mutation factor acquisition unit is used to acquire the current mutation factor based on the current time series standard deviation, the current time series maximum value, and the current time series minimum value. A thermal deformation coefficient acquisition unit is used to acquire the initial temperature distribution vector and the real-time temperature distribution vector of the tool, and to acquire the thermal deformation coefficient based on the initial temperature distribution vector and the real-time temperature distribution vector. The basic risk probability acquisition unit is used to acquire the basic collision risk probability based on the current mutation factor, thermal deformation coefficient, and the compensated and corrected trend term.

9. A protection system for a CNC machining center as described in claim 8, characterized in that, Also includes: The seasonal item acquisition module is used to decompose the vibration signal to obtain the seasonal item; The final risk probability acquisition module is used to compensate and correct the basic collision risk probability based on the seasonal term to obtain the final collision risk probability.

10. A protection system for a CNC machining center as described in claim 9, characterized in that, Also includes: The risk adjustment term acquisition module is used to acquire the sensitivity adjustment factor, construct a multidimensional risk feature vector based on the current mutation factor, thermal deformation coefficient and the compensated and corrected trend term, and map the product of the sensitivity adjustment factor and the multidimensional risk feature vector through the Sigmoid function to acquire the risk adjustment term. The maximum possible displacement acquisition module is used to obtain the maximum possible displacement of the tool within a future preset time period based on the tool's instantaneous feed rate, tool acceleration, and system control cycle. The real-time Euclidean distance acquisition module is used to acquire the machine tool rigidity coefficient and the real-time Euclidean distance between the tool and the nearest obstacle; An adaptive safety distance acquisition module is used to acquire an adaptive safety distance based on the ratio between the product of the machine tool rigidity coefficient, the risk adjustment term, and the real-time Euclidean distance and the maximum possible displacement. The avoidance strategy execution module is used to execute a preset avoidance strategy when the adaptive safety distance is less than a preset distance safety threshold.

Images